If replacing with a ceramic you need to go for about twice the capacitance IIRC, or go with an MLCC. FWIW I find the MLCC replacement pointlessly pricy, so I just replace them with electrolytics and haven't had an issue yet.

Why not use Tantalums - again?
Tantalums out of current production by renowned brands like AVX are far more stable and rugged than the ones manufactured in the 80ies. They are low impedance / inductance by design just like a top tier electrolytic. Tantalums also won´t dry out.
ESR at highest frequencies isn´t an issue on a XT system that runs at "low frequencies" compared to current systems.
Plus you keep the original looks 😀

(Perfect solution ESR-wise would probably be a chip capacitor soldered on the backside of the PCB right between the landings for the old capacitors - or new Tantalums and an additional smaller (10 - 100nF) chip capacitor on the soldering side for each Tantalum for optimizing ESR at low frequencies <10KHz. This would also preserve the original looks.)

clownwolf wrote on 2023-07-20, 07:18:

something like this:
https://www.digikey.com/en/products/detail/td … 06MRT06/7384735

Excellent choice for 5V bus.
For 12V, I`lld recommend capacitor in big case, like 1210.
PS. Molded case its MLCC with solder wire pins and little epoxy

majestyk wrote on 2023-07-20, 07:52:

Tantalums and an additional smaller (10 - 100nF) chip capacitor on the soldering side for each Tantalum.

Absolutely pointless.
100nF is a number that became an "urban legend" at a time when there were no high-capacity ceramic capacitors.
It makes sense to set the capacitance to 1-10% of the capacitance of the electrolytic / tantalum capacitor.

100nF is a local RF isolation right next to the pins of the chip.

majestyk wrote on 2023-07-20, 07:52:

Why not use Tantalums - again?

Because they explode, sooner or later. Particularly when they are decoupling/filtering caps on an AT power rail. The -12V ones are the worst in my experience. I had one explode in my face only two nights ago. I'd rather replace them with something that will almost always fail open then something that can fail short or even worse damage other things around it. Including me.. 😅

appiah4 wrote on 2023-07-20, 08:38:

majestyk wrote on 2023-07-20, 07:52:

Why not use Tantalums - again?

Because they explode, sooner or later. Particularly when they are decoupling/filtering caps on an AT power rail. The -12V ones are the worst in my experience. I had one explode in my face only two nights ago. I'd rather replace them with something that will almost always fail open then something that can fail short or even worse damage other things around it. Including me.. 😅

Do you think those fulfilling MIL standards and everything are still going to explode?
https://catalogs.kyocera-avx.com/HighReliabilityTantalum.pdf

shevalier wrote on 2023-07-20, 08:33:

Absolutely pointless. 100nF is a number that became an "urban legend" at a time when there were no high-capacity ceramic capacit […]
Show full quote

majestyk wrote on 2023-07-20, 07:52:

Tantalums and an additional smaller (10 - 100nF) chip capacitor on the soldering side for each Tantalum.

Absolutely pointless.
100nF is a number that became an "urban legend" at a time when there were no high-capacity ceramic capacitors.
It makes sense to set the capacitance to 1-10% of the capacitance of the electrolytic / tantalum capacitor.

100nF is a local RF isolation right next to the pins of the chip.

But would 10 - 100nF hurt? They are often more suitable to bridge the distance for 5mm pitch ( or even 2.5mm).

majestyk wrote on 2023-07-20, 08:47:

appiah4 wrote on 2023-07-20, 08:38:

majestyk wrote on 2023-07-20, 07:52:

Why not use Tantalums - again?

Because they explode, sooner or later. Particularly when they are decoupling/filtering caps on an AT power rail. The -12V ones are the worst in my experience. I had one explode in my face only two nights ago. I'd rather replace them with something that will almost always fail open then something that can fail short or even worse damage other things around it. Including me.. 😅

Do you think those fulfilling MIL standards and everything are still going to explode?
https://catalogs.kyocera-avx.com/HighReliabilityTantalum.pdf

Probably not, but then do you think they are less expensive than a bunch of Koshin low ESR electrolytics that just do the same job?

Id prefer not to use tantalums, but I am willing to re-use them if I pull them out of a fairly new source. This is because I power on my XT boards only every 10 years or so, since the only use-case I have for them right now is Ultima 4-6.
In fact I didn't even use the regular wet electrolytics for my 5160 board, since I leave them in storage for so long. I used solid polymer electrolytics that I pulled from donors.

shevalier wrote on 2023-07-20, 08:18:

Excellent choice for 5V bus. For 12V, I`lld recommend capacitor in big case, like 1210. PS. Molded case its MLCC with solder wir […]
Show full quote

clownwolf wrote on 2023-07-20, 07:18:

something like this:
https://www.digikey.com/en/products/detail/td … 06MRT06/7384735

Excellent choice for 5V bus.
For 12V, I`lld recommend capacitor in big case, like 1210.
PS. Molded case its MLCC with solder wire pins and little epoxy

What do you mean by "big case". Do you mean the physical dimensions? If yes, then the chemi-con 50v is the largest through hole from Digikey under ~$2. Its regular ceramic, Digikey's MLCCs are all surface mount.
https://www.digikey.com/en/products/detail/ch … 55A0T00/1850869

Its expensive... so I will have to buy bulk 100 50 pcs to ironically save money. But I am fine with spending on high-grade capacitors. I will have to use them on the 5v rails as well, possibly re-cap tantalums on graphics cards and re-cap my IBM 5160 again.

clownwolf wrote on 2023-07-20, 10:24:

What do you mean by "big case". Do you mean the physical dimensions? If yes, then the chemi-con 50v is the largest through hole from Digikey under ~$2. Its regular ceramic, Digikey's MLCCs are all surface mount.
https://www.digikey.com/en/products/detail/ch … 55A0T00/1850869

Its expensive... so I will have to buy bulk 100

Yes.
A modern ceramic capacitor with leads is an MLCC with soldered leads.
For X5 / X7 capacitors with the use of bias voltage, the capacitance drops significantly.
The smaller the case, the greater the drop.
0805 10u* 25V at 12VDC will be 2-3uF.
1210 10u* 25V at 12VDC will have 6-7uF.
For a 5V bus, those that were in the first message will do.
Usually there are 3-5 capacitors in the 12V bus.
PS.
MLCC is the internal structure of a capacitor, just like a film or foil.
Leaded or SMD - type of case.

To all
the great tool from Kemet
https://ksim3.kemet.com/capacitor-simulation

I am now thinking about giving Tantalums another chance. I was looking at 50v rated tantalums, and they seem to be about the same price as the 50v rated MLCCs.

Tantalums seems to be a better fit for retro use, since they do not age. Also, using 50v instead of 16v-rated tantalums would technically reduce the chance of explosions on the 12v rail capacitor filters right?

No using 50v caps instead of 16v won't make a difference if one shorts, it will still go "POOF" 🤣

appiah4 wrote on 2023-07-20, 08:38:

majestyk wrote on 2023-07-20, 07:52:

Why not use Tantalums - again?

Because they explode, sooner or later. Particularly when they are decoupling/filtering caps on an AT power rail. The -12V ones are the worst in my experience. I had one explode in my face only two nights ago. I'd rather replace them with something that will almost always fail open then something that can fail short or even worse damage other things around it. Including me.. 😅

It depends.. Not necessarily they do go boom.
As far as I was told once, a problem with Tantal condensers (dry) is that they can loose or "forget" their polarity after a long time of inactivity (years in storage).
If they're powered on first time after such a long time, they must try to reformat themselves.
If the "shock" to them is too great, they break. It's as if they were exposed to reverse-polarity, essentially.
So it's not necessarily required to replace all old tantals. If they were running stable for a few hours, the chances of them breaking decreases.

Jo22 wrote on 2023-07-23, 01:48:

It depends.. Not necessarily they do go boom. As far as I was told once, a problem with Tantal condensers (dry) is that they can […]
Show full quote

It depends.. Not necessarily they do go boom.
As far as I was told once, a problem with Tantal condensers (dry) is that they can loose or "forget" their polarity after a long time of inactivity (years in storage).
If they're powered on first time after such a long time, they must try to reformat themselves.
If the "shock" to them is too great, they break. It's as if they were exposed to reverse-polarity, essentially.
So it's not necessarily required to replace all old tantals. If they were running stable for a few hours, the chances of them breaking decreases.

I have known that too ! Also same with electolytics, they need to powered on every so often or internally weaken.

The risk of Tantalums shorting / exploding increases dramatically when they are operated near their nominal voltages. This is why on mainboards in almost all cases the tantalums in the +12V and -12V rails fail, when they have a nominal voltage of 16V.
This can be avoided by choosing the next higher nominal voltage like 25V or 35V.

This effect has already been observed in the 60ies - 80ies in audio /HiFi equipment.

clownwolf wrote on 2023-07-22, 21:56:

I am now thinking about giving Tantalums another chance. I was looking at 50v rated tantalums, and they seem to be about the same price as the 50v rated MLCCs.

Tantalums seems to be a better fit for retro use, since they do not age. Also, using 50v instead of 16v-rated tantalums would technically reduce the chance of explosions on the 12v rail capacitor filters right?

https://www.avx.com/docs/techinfo/VoltageDera … mCapacitors.pdf
https://studylib.net/doc/18572467/application … alum-capacitors
The failure rate of a low impedance circuit at 0.1 Ohm/V is about five times greater than that of a 1 Ohm/V circuit. To curtail
this higher failure rate, Tantalum capacitors used in low impedance circuits, such as filters for power supplies (particularly
switching power supplies), or for noise by-passing, require that operating voltage be derated to less than half of the rated
voltage. Actually, 1/3 of the rated voltage is recommended.

appiah4 wrote on 2023-07-20, 08:38:

majestyk wrote on 2023-07-20, 07:52:

Why not use Tantalums - again?

Because they explode, sooner or later.

Yea, maybe 30-50 years in the future from the replacement date.

wiretap wrote on 2023-07-23, 15:04:

Yea, maybe 30-50 years in the future from the replacement date.

No, boards (at the time of their relevance) with previously exploded capacitors have long been recycled.
Now it's the turn of the lucky survivors.

shevalier wrote on 2023-07-23, 15:40:

wiretap wrote on 2023-07-23, 15:04:

Yea, maybe 30-50 years in the future from the replacement date.

No, boards (at the time of their relevance) with previously exploded capacitors have long been recycled.
Now it's the turn of the lucky survivors.

Manufacturing defect in a capacitor or another fault causing it blow. I still run hundreds of boards at work with tantalum capacitors in the 30-40yr old range that have yet to blow one. They're definitely an extremely reliable design. I don't hesitate when replacing them with the same thing on my vintage computing parts because I know they'll make it another several decades. Manufacturing now is leaps and bounds better for those parts than it was in the 70's/80's/90's. The reject rates on the assembly line is almost 10 fold lower.

wiretap wrote on 2023-07-23, 20:12:

shevalier wrote on 2023-07-23, 15:40:

wiretap wrote on 2023-07-23, 15:04:

Yea, maybe 30-50 years in the future from the replacement date.

No, boards (at the time of their relevance) with previously exploded capacitors have long been recycled.
Now it's the turn of the lucky survivors.

Manufacturing defect in a capacitor or another fault causing it blow. I still run hundreds of boards at work with tantalum capacitors in the 30-40yr old range that have yet to blow one. They're definitely an extremely reliable design. I don't hesitate when replacing them with the same thing on my vintage computing parts because I know they'll make it another several decades. Manufacturing now is leaps and bounds better for those parts than it was in the 70's/80's/90's. The reject rates on the assembly line is almost 10 fold lower.

Is that really true? I'd love to see some sources, if there are any.
Because, the global picture looks different to me.
Electronics are getting more and more cheap, which to me makes it hard to believe in any kind of progress.
The electronics parts my father had bought in the 70s do have golden legs, I remember.
They're from known-good sources too, since Chinese suppliers didn't exist yet.
The ICs he had (has) do have ceramic packaging and are made using TTL and NMOS process, both which have more "oomph" than CMOS designs.
They're all 5v tolerant, as well. And produced in a bigger manufacturing process, meaning they're more though.